Merge branch 'develop'

.gitignore

@@ -37,3 +37,4 @@ bin/
 
 # Miscellaneous
 .vscode/
+external/

CMakeLists.txt

@@ -1,4 +1,5 @@
 
+
 # Specify the minimum version of CMake to use.
 # CMake can have different behaviors (policies) based on its version used.
 cmake_minimum_required(VERSION 3.14)
@@ -7,14 +8,14 @@ cmake_minimum_required(VERSION 3.14)
 # Name the project, its version, and languages used in it.
 set(PROJECT_NAME cuttlefish)
 project(${PROJECT_NAME}
-        VERSION 1.0.0
+        VERSION 2.0.0
         LANGUAGES CXX C
         )
 
 
 # Fix language standards, and set hard requirements for such.
 # All targets defined from this point onward will pick up these requirements.
-set(CMAKE_CXX_STANDARD 14)
+set(CMAKE_CXX_STANDARD 17)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
 
 set(CMAKE_C_STANDARD 11)
@@ -38,20 +39,139 @@ set(OPTIMIZE_FLAGS -funroll-loops)
 
 
 # Add the required preprocessor definitions (`#define`s) to pass on.
-add_definitions(-D__STDC_FORMAT_MACROS -DSPDLOG_FMT_EXTERNAL_HO -DFMT_HEADER_ONLY -DXXH_INLINE_ALL)
+# TODO: find out what are `__STDC_FORMAT_MACROS` and `SPDLOG_FMT_EXTERNAL_HO` for.
+add_compile_definitions(__STDC_FORMAT_MACROS SPDLOG_FMT_EXTERNAL_HO FMT_HEADER_ONLY XXH_INLINE_ALL)
+
+add_compile_definitions(PROJECT_VERSION=${CMAKE_PROJECT_VERSION})
+
 if(INSTANCE_COUNT)
-    add_definitions(-DINSTANCE_COUNT=${INSTANCE_COUNT})
+    add_compile_definitions(INSTANCE_COUNT=${INSTANCE_COUNT})
+endif()
+
+if(CF_VALIDATION_MODE)
+    add_compile_definitions(CF_VALIDATION_MODE)
+endif()
+
+if(CF_DEVELOP_MODE)
+    add_compile_definitions(CF_DEVELOP_MODE)
+endif()
+
+
+# Here, we have some platform-specific considerations
+# of which we must take care.
+if(CMAKE_SYSTEM_NAME STREQUAL "Linux")
+    execute_process(
+        COMMAND getconf LEVEL1_DCACHE_LINESIZE
+        COMMAND tr -d '\n'
+        OUTPUT_VARIABLE L1_CACHE_LINE_SIZE
+    )
+elseif(CMAKE_SYSTEM_NAME STREQUAL "Darwin")
+    # OSX has `getconf`, but doesn't report `LEVEL1_DCACHE_LINESIZE`;
+    # so we instead use `sysctl` for the corresponding variable.
+    execute_process(
+        COMMAND sysctl machdep.cpu.cache.linesize
+        COMMAND awk "{print $2}"
+        COMMAND tr -d '\n'
+        OUTPUT_VARIABLE L1_CACHE_LINE_SIZE
+    )
+
+    # Later on, jemalloc will complain if the C compiler 
+    # hasn't been properly set.
+    if(NOT CONDA_BUILD)
+        set(CMAKE_C_COMPILER clang)
+        set(CMAKE_CXX_COMPILER clang++)
+    endif()
 endif()
 
+add_compile_definitions(L1_CACHE_LINE_SIZE=${L1_CACHE_LINE_SIZE})
+
 
 # Search the file system for the appropriate threads package for this platform, and then set
 # the `CMAKE_THREAD_LIBS_INIT` variable (and some other variables as well).
 find_package(Threads REQUIRED)  # The threads package is required for the BBHash library used in the project.
 set(THREADS_PREFER_PTHREAD_FLAG TRUE)   # The BBHash library uses `pthread`.
 
 # Search and load setting for the `zlib` library. The library is required to seamlessly adapt
-# the `kseq` library to gzipped compressed files.
-find_package(ZLIB REQUIRED)
+# the `kseq` and the `kmc` libraries to gzip-compressed files.
+include(FindZLIB)
+if(NOT ZLIB_FOUND)
+    message(FATAL_ERROR "zlib (https://zlib.net/) is required. Aborting.")
+endif()
+
+# Search and load setting for the `bzip2` library. It is required to seamlessly adapt the
+# `kmc` library to bzip-compressed files.
+include(FindBZip2)
+if(NOT BZIP2_FOUND)
+    message(FATAL_ERROR "bzip2 (https://sourceware.org/bzip2/) is required. Aborting.")
+endif()
+
+
+# Set path for modules required to search for existing packages in the system.
+set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_SOURCE_DIR}/cmake/Modules/")
+
+
+# Module required to download and install external projects.
+include(ExternalProject)
+
+set(EXT_LIB ${CMAKE_SOURCE_DIR}/external/lib/)
+set(EXT_INCLUDE ${CMAKE_SOURCE_DIR}/external/include/)
+
+file(MAKE_DIRECTORY ${EXT_LIB} ${EXT_INCLUDE})
+
+
+# Prepare the `jemalloc` library. It provides scalable concurrency support and better avoidance
+# of fragmentation.
+set(MALLOC_LIB "")
+set(JEMALLOC_MIN_VERSION "5.2.1")
+find_package(Jemalloc ${JEMALLOC_MIN_VERSION})
+if(JEMALLOC_FOUND)
+    message("Found jemalloc (v${JEMALLOC_VERSION}) in the system")
+    set(MALLOC_LIB ${JEMALLOC_LIBRARIES})
+else()
+    message("Build system will fetch and install jemalloc")
+    ExternalProject_Add(prj_jemalloc
+        DOWNLOAD_DIR        ${CMAKE_SOURCE_DIR}/external
+        DOWNLOAD_COMMAND    curl -k -L https://github.com/jemalloc/jemalloc/archive/5.2.1.tar.gz -o jemalloc-5.2.1.tar.gz &&
+                            tar -xzf jemalloc-5.2.1.tar.gz &&
+                            rm jemalloc-5.2.1.tar.gz
+
+        SOURCE_DIR          ${CMAKE_SOURCE_DIR}/external/jemalloc-5.2.1
+        BUILD_IN_SOURCE     TRUE
+        INSTALL_DIR         ${CMAKE_SOURCE_DIR}/external
+        CONFIGURE_COMMAND   sh -c "CC=${CMAKE_C_COMPILER} ./autogen.sh --disable-debug --enable-static --prefix=<INSTALL_DIR> --silent"
+        INSTALL_COMMAND     cp lib/libjemalloc.a ${EXT_LIB}
+    )
+
+    set(MALLOC_LIB ${EXT_LIB}/libjemalloc.a)
+endif()
+
+add_library(jemalloc STATIC IMPORTED)
+set_target_properties(jemalloc PROPERTIES IMPORTED_LOCATION ${MALLOC_LIB})
+if(NOT JEMALLOC_FOUND)
+    add_dependencies(jemalloc prj_jemalloc)
+endif()
+
+
+# Prepare the `kmc` library — required by the Cuttlefish algorithm implementation.
+message("Build system will fetch and install KMC3")
+ExternalProject_Add(prj_kmc
+    DOWNLOAD_DIR        ${CMAKE_SOURCE_DIR}/external
+    DOWNLOAD_COMMAND    curl -k -L https://github.com/refresh-bio/KMC/archive/refs/tags/v3.2.1.tar.gz -o KMC-3.2.1.tar.gz &&
+                        tar -xzf KMC-3.2.1.tar.gz &&
+                        rm KMC-3.2.1.tar.gz
+    PATCH_COMMAND       patch --strip 1 < ${CMAKE_SOURCE_DIR}/patches/kmc_patch.diff
+    SOURCE_DIR          ${CMAKE_SOURCE_DIR}/external/KMC-3.2.1
+    BUILD_IN_SOURCE     TRUE
+    INSTALL_DIR         ${CMAKE_SOURCE_DIR}/external/
+    CONFIGURE_COMMAND   ""
+    BUILD_COMMAND       make -j CC=${CMAKE_CXX_COMPILER}
+    INSTALL_COMMAND     cp bin/libkmc_core.a ${EXT_LIB} &&
+                        cp include/kmc_runner.h ${EXT_INCLUDE}
+)
+
+add_library(kmc STATIC IMPORTED)
+set_target_properties(kmc PROPERTIES IMPORTED_LOCATION ${EXT_LIB}/libkmc_core.a)
+add_dependencies(kmc prj_kmc)
 
 
 # The `Debug` configuration optimizes the program for debugging and enables full debug information.
@@ -67,5 +187,11 @@ set(CMAKE_C_FLAGS_DEBUG "-g")
 set(CMAKE_C_FLAGS_RELEASE "-O3")
 
 
+# Help `conda` build for OSX through circumventing some of its old SDK-based checks.
+if(CONDA_BUILD)
+    set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -D_LIBCPP_DISABLE_AVAILABILITY")
+endif()
+
+
 # Add subdirectory `src` to the build; CMake will open `src/CMakeLists.txt` for such.
 add_subdirectory(src)